Transformers v5 (#2408)

Commit

7 days ago

Transformers v5 (#2408) * Uses AutoModelForImageTextToText instead of AutoModelForVision2Seq * image-to-text pipelines is deprecated in v5